Latency of Forge Custom UI bridge/resolver

Hey everyone,

I am seeing very long latencies when calling Forge functions from Custom UI, especially when comparing to a Connect app. I have reduced my code to the bare minimum, but still have not seen any call taking less than 1,800ms which seems really long.

Here’s my stripped down resolver code:

resolver.define('get-all-data', async ({ context }) => {
  return [];

I can trace the call in Safari’s developer console, a typical call would look like this:

Bildschirmfoto 2021-04-08 um 16.41.34

Meaning that (in this case) 3842.1 ms are spend waiting for a response which then takes 0.3 ms to load. Our Jira instance seems to be in AWS eu-central region, which is geographically close to where I am.

When I make similar calls in a connect application, I see much smaller latencies. As in orders of magnitude smaller. Here’s the trace of a comparable call to a (production) connect backend (in AWS us-east, so a bit further away):

Bildschirmfoto 2021-04-08 um 16.59.38

I understand that a lot of factors can affect latency, but a 20-36x difference is a lot. The high latency would make Forge an unfeasible choice for building our app.

Has anyone had a similar problem and how did you work around it?